Update method of hashing the SignedTxRequest in the circuits.

[spalladino]

As part of this discussion, we agreed that we need a way to reconstruct tx hashes for private and public txs from the L2 block. Given it's possible it'll change in the near future, rather than passing in a new field altogether with the tx hash, we'll be using the first nullifier of each tx to store the hash. We'll also alter how we calculate tx hashes: for all txs, public or private, they'll be the hash of the signed tx request.

This means we need to make the following changes:

• In the Tx object in the types package, update how tx hashes are calculated. The hash should always be obtained by pedersen hashing the signed tx request (check if we need a new generator point). For the sake of simplicity, we'll disallow txs that have both a private and a public signed tx request.
• In both the private and public kernel circuits, calculate the tx hash from the signed tx request, ordering the fields in the same way as in the client.
• In the archiver/sequencer, when processing a new L2 block that was observed on L1, just use the first nullifier as the tx hash to figure out which txs need to be removed from the P2P pool.

For reference, the hash of a tx should be calculated by concatenating the following fields in the following order, and pedersen hashing them:

from
to
functionData
args
nonce
txContext
chainId
signature.r
signature.s

[spalladino]

Broken down into:

• Update components to take tx hash from first tx nullifer (settled txs) #374
• Private kernel circuit should calculate the hash of the SignedTxRequest and emit the value as the first nullifier #375
• Update public circuit to calculate tx hash #376
• Public kernel circuit should calculate the hash of the SignedTxRequest and emit the value as the first nullifier #377

[dbanks12]

Note there is now a third entry in the signature struct signature.v that might need to be added in this list.